Comparative analysis of cast-in-place post-tensioned and steel–concrete composite bridge bent caps

Structural concrete prefabricated systems have been successfully used with the well-known advantages of construction time reduction and the number of operations to be performed in situ, improved quality control, work zone safety improvement and lower environmental impact. Unlu [1] proposed various precast concrete systems for the construction of abutments and bent caps. Billington et al. [2] proposed concrete precast systems for the construction of single- and double-column inverted T hammerhead pier caps with connection details. Matsumoto et al. [3] developed a precast concrete system for pile bent caps for non-seismic regions with several connection details. Other systems have been applied successfully. For example, precast post-tensioned supported bent caps with an inverted T-section with spans up to 17.68 m were used for the U-section bridges in the project mentioned herein. However, there is a practical limit for the span lengths due to the self-weight that can impact safe lifting and placement.

The evolution of pier cap design is a consequence of the intensive use of bridges caused by the changes in the transportation infrastructures [4]. In general, abutment design has been extensively analyzed [4], particularly for integral abutments. Barr et al. [5] investigated the spalling of an integral abutment bridge in Salt Lake City, UT, through monitoring and finite element modeling. Nikravan [6] studied the behavior of integral bridges subjected to temperature variations using 3D finite element models. However, pier choices are generally limited to CIP concrete supported either by one hammerhead bent cap or more than one column and pile bent cap [7]. Some special types (straddle bent caps) are briefly addressed elsewhere [8]. Public agencies have guidelines on the design of the most common bent cap types, such as rectangular column bent caps [9, 10], inverted T column bent caps [9, 10] and hammerhead bent caps [11]. However, comparative design/construction guidance for alternates is lacking.

Nicholas et al. [12] compared the reinforcing requirements of the strength design approach for flexure and shear and the strut-and-tie model method in their application to hammerhead pier design. Pereira [13] analyzed the behavior of hammerhead piers reinforced with T-headed bars and different levels of prestressing in the cantilever ends by static load testing. Kerley [14] developed finite element models to analyze the behavior of the connections between reinforced concrete bent caps and cast-in-shell piles under lateral loads. Bracci et al. [15] experimentally studied unexpected cracks in the cantilevered regions of concrete bent caps under service loads. The flexural cracking was related to the stress levels in longitudinal reinforcement below the service stress limits. Billington [16] experimentally studied the structural behavior of two-span continuous bent caps with different levels of prestress. Several past studies involved integral piers. For example, Ales [17] developed a connection detail between an integral steel cap girder and concrete piers, while Wassef et al. [18] developed specifications and design examples for integral steel box-beam pier caps. Denio et al. [19] studied the shear strength and reinforcing details for connections between steel bent caps and concrete piers through static load testing.

It is apparent that a comparative study of the two alternative bent caps considered herein has not been undertaken in the past. The cast-in-place post-tensioned cap is a more traditional approach, while the steel–concrete composite cap is a more innovative solution. Possible cost and time savings from the composite cap utilization, if demonstrated, would allow the possible consideration and adoption of this type and other types of innovative bent cap solutions.

The IH-635 managed lane construction project is located in North Dallas, Texas, and involves both IH-35E and IH-635 freeways (Fig. 1). Started in early 2011, it was completed and opened to traffic in summer of 2015. It has dramatically increased the traffic capacity of this important Dallas corridor, consisting of four to six general purpose lanes, two managed lanes and two to three frontage roads in each direction. A new loop 12–IH-635 westbound direct connector was built, needing the construction of a new bridge (Bridge 4) to crossover IH-35E northbound, as seen in Fig. 2. This is a major traffic corridor, supporting high traffic volume, particularly during rush hours. The construction of a new bridge using conventional design and construction techniques on this corridor could result in significant lane closings, impacting traffic and overall economic activity.

The originally conceived and designed post-tensioned cast-in-place (CIP) concrete bent cap could not be built without a lengthy closure of IH-35E for the required formwork and concrete curing times. As a result, the design team considered as alternate an innovative prefabricated bent cap solution simply supported on columns and straddling the highway underneath that could limit the traffic impact. With bent spans varying from 22.7 to 26.2 m (Fig. 3), the two alternative precast designs considered were: (1) a precast prestressed concrete bent cap and (2) a composite steel–concrete bent cap, consisting of a rectangular-shaped steel section with a top CIP concrete slab. The two alternatives reduced the various constructability issues associated with the original CIP concept that would include time-consuming post-tensioning of tendons at an elevated position. The second alternative involves simply lifting and placing the steel–concrete section once the concrete slab reached enough strength to support its self-weight.

Bridge 4 (Fig. 3) is the focus of this paper. With a total length of 304.8 m, the bridge consists of nine spans with three Tx54 standard concrete girders [20, 21] per span, with the exception of span three that has three steel plate girders. The future lateral extension of the bridge with two new girders per span was considered in the design. Spans 6–9 and bent caps 7–9 were built over the IH-35E roadway. The two precast alternatives discussed previously for bent 7 (22.7 m span length) in Bridge 4 were analyzed herein for structural behavior, cost and construction schedule. The characteristics of the two alternatives are presented in Table 1. They were considered as simply supported on two CIP 1.83-m-diameter concrete columns. Elastomeric bearing pads and a shear key were employed to ensure the end conditions. Bent 7 carries the loads from spans 6 and 7 (Figs. 4, 5).

The current bridge construction market conditions, with increasing competitiveness among the construction companies, have made the design optimization key for large transportation projects. Therefore, a theoretical comparative study between the two alternatives in terms of structural behavior, cost-efficiency and construction time was undertaken and presented in this paper [22].

The FEM software ABAQUS [23] includes several material models that were considered herein. For concrete in both alternatives, the concrete damage plasticity (CDP) model was used in this study, intended for brittle materials with the possibility of establishing failure criteria by damage parameters. Both tension and compression are treated as linear elastic–plastic, and tension stiffening is used to model the tension post-failure behavior and reinforcement interaction. This is the most realistic concrete model in ABAQUS. The concrete was modeled using a 3D deformable tetrahedral homogeneous solid element, due to its ability to better adapt to any member geometry. Although mesh refinement needed is usually smaller if quadratic elements are used, only linear elements were utilized in this study because they are more accurate for plastic behavior. For concrete, AASHTO provides a Poisson’s ratio of 0.2 [24]. The CDP default parameters were taken herein as follows: dilatation angle 36°, eccentricity 0.1 and viscosity parameter 0 [25]. In the absence of experimental data, the approach suggested by Wight and MacGregor [26] for the concrete compressive behavior was used herein, valid for concrete strengths from 13.8 to 124 MPa. The needed damage parameters were adopted from Birtel and Mark [27]. A maximum stress and the corresponding strain must be defined as a limit for total compression crushing failure of the concrete and were taken as 0.5 and 0.3 times the concrete compressive strength, respectively [28]. The resulting concrete compressive stress–strain model is presented in Fig. 6.

A simplified tensile concrete stress–strain relationship, as shown in Fig. 7 [28], was used herein with an elastic behavior up to the modulus of rupture. The subsequent discharge branch that defines the plastic behavior of the material is approximated with three line segments.

The regular steel rebars in concrete for both alternatives were modeled as 3D wire elements with truss sections (capable of developing only axial stresses) with a cross-sectional area equal to the rebar area. To guarantee the deformation compatibility and proper stress transmission between concrete and rebar, the reinforcement was embedded into the concrete section (host region). By doing so, the degrees of freedom of the truss elements were constrained by the adjacent concrete solid elements. The Grade 60 reinforcing steel was modeled with an elastic–perfectly plastic behavior and a yield strength of 413.7 MPa. A steel modulus of elasticity of 200,000 MPa with a Poisson’s ratio of 0.3 was assumed.

The prestressing steel for the post-tensioned alternative 1 was also modeled as elastic–plastic, with the following properties: modulus of elasticity of 196,500 MPa, Poisson’s ratio of 0.3, ultimate strength of 1,861 MPa, yield strength of 1,675 MPa, ultimate tensile strain of 0.035, 31 strands of 15 mm diameter each and nominal area of 4,340 mm² per tendon.

For alternative 2, the structural steel stress–strain curve was also assumed to be elastic–perfectly plastic. The A709 Grade 50 steel had a modulus of elasticity of 200 GPa, Poisson’s ratio of 0.3 and yield strength of 345 MPa. The steel section was modeled using 3D deformable shell elements. The top flange was tied to the bottom surface of the concrete slab to guarantee deformation compatibility. Using the tie interaction, the nodal degrees of freedom in the slab surface were constrained by those on the cap surface. Three types of stiffeners were included: diaphragms at support locations, girder stiffeners at girder locations and interior stiffeners distributed along the length of the member. They were modeled with 3D deformable homogeneous shells. Proper interaction between the cap section and the stiffeners was achieved through merging.

The two alternate models were subjected to AASHTO HL-93 loading [24]. The loads applied to the bent caps consisted of the following: (1) girder reactions at supports for the controlling load combination (strength I) obtained by considering the superstructure self-weight (girder, deck, curbs and barriers), the lane load and the truck or tandem load; and (2) factored bent cap self-weight for the critical load combination (strength I). The cap concrete and the structural steel unit weights were assumed as 23 and 77.8 kN/m³, respectively. The girder reactions were determined with the software PGSuper [20, 21] by modeling the two bridge spans that load the pier cap. This is a windows-based software for the design, analysis and load rating of multi-span precast–prestressed bridge girders in accordance with the AASHTO LRFD specifications. Table 2 summarizes the girder reactions on the bent cap. Cap models supported 10 girders (five from the backward span and five from the forward span).

Simple support conditions on the piers were assumed for the bent caps and were achieved through the release of support constraints as needed. Releasing the rotation at supports minimized the moment transferred to the columns.

Figure 8 shows the prestressing tendon profile view of the alternate 1 bent cap. The effect of the post-tensioning tendons was modeled as a combination of homogeneous axial compressive stress and a vertical force from the draped tendons, from the load-balancing method. Because the contribution of prestressing steel to the stiffness of the section is minimum [29], actual prestressing tendons were not included in the FEM. The compressive force applied at the ends was determined by deducting the prestress losses from the jacking force. The load-balancing force was obtained by imposing equilibrium at the tensioned tendon.

The prestress force at jacking per tendon was 4,328 kN. The effective prestress was obtained using AASHTO LRFD procedure [24]. The total loss of prestress was determined as the sum of the short-term losses (elastic shortening of 20 MPa, friction loss of 54 MPa, anchorage set loss of 76 MPa) and long-term losses (shrinkage/creep, steel relaxation) of 249 MPa, resulting in a total loss of 399 MPa. The approximate method given by AASHTO LRFD, article 5.9.5.3, was used herein to find the long-term losses, with an assumed relative humidity of 70%. The resulting effective prestress was 1,144 MPa. A total prestress loss of 399 MPa was obtained, resulting in the net long-term prestress force of 4,328 kN per tendon.

Table 3 shows the mesh details used herein. Figure 9 shows the alternate 1 model with 127 mm mesh size. Computer run times were found to be exceedingly high for mesh sizes below the minimum mesh size shown.

Simplified elastic hand calculations performed herein were used to validate the FEM. The contribution of the longitudinal steel to the section strength was neglected. The solution obtained implied close match between the FEM and the hand-calculated stresses. The concrete elements remained within the elastic limits for both top and bottom fiber stresses. As can be seen, the FEM adequately simulated the structural behavior of the bent cap (Fig. 10).

The plan and section views for the alternative 2 composite cap are shown in Figs. 11 and 12, respectively. Figure 13 shows the composite cap meshed view with 76 mm mesh size, and Fig. 14 shows the bending stresses distribution in the steel section part.

Accurately determining the cost of a construction project is a difficult task that needs experience, engineering judgment and scientific principles [30]. Different methods are available for the estimator depending on the project stage and information available, and accuracy needed [31]. The direct cost of the pier cap alternatives considered herein was estimated for comparison purposes.

The following materials would be needed for the construction of the post-tensioned bent cap: concrete with a 28-day compressive strength of 41.4 MPa; grade 60 reinforcing steel; grade 270 low relaxation prestressing steel; and other materials, such as grout, plastic ducts and end anchorages. The construction process would consist of the following activities: (1) shoring structure placement until the cap reaches sufficient strength; (2) long-term closure of traffic in the roadway below the cap until the end of the cap construction process; (3) form work; (4) rebar cage fabrication and placement performed simultaneously with previous activities; (5) placement of additional elements for prestressing tendons (ducts and end anchorages), simultaneously with the previous tasks; (6) concrete casting with sufficient curing time to achieve the minimum required strength to perform the next two activities; (7) tendon placement, stressing and duct grouting; and (8) form work and shoring structure removal.

For typical moist-cured concrete with type I cement, the variation of strength with time is defined in Eq. 1 [26]:where \(f_{\text{ci}}^{\prime }\) denotes concrete strength at time t; t curing time (days); and \(f_{c}^{\prime }\) nominal 28-day concrete strength.

Considering the 28-day concrete compressive strength of 41.4 MPa and a required strength at post-tension jacking of 34.5 MPa, a minimum duration of 12 days after concrete casting would be needed before the tendons could be stressed.

The following materials would be needed for the construction of the composite bent cap: concrete with a nominal compressive strength of 41.4 MPa, Grade 60 reinforcing steel and A709 Grade 50 structural steel. The construction process would require a lower number of activities than the post-tensioned cap process, as follows: (1) concrete slab form work. The steel section is placed on ground and the slab formwork is placed on it, also at ground level; (2) slab rebar placement also at ground level; (3) slab concrete casting also at ground level. Enough curing time has to be provided prior to removing the form work. (4) Slab form work removal; and (5) lifting and placement of the composite cap in place on already built CIP columns (Fig. 15).

Considering a minimum required concrete strength of 20.7 MPa for lifting, from Eq. 1, the lifting and placement of the composite cap could not take place before 4 days after slab casting. IH-35E highway would have to be closed to traffic during Task 5. Because of the short time required to place the cap on its final position, only one night closure of IH-35E would be needed.

The cost of a project development can be obtained as the summation of the following costs for each of the materials and activities: First is direct cost of material, labor and equipment. This cost, provided that the construction process is optimized, should not significantly vary with the construction company performing the work. And second is indirect cost of traffic closure and detour.

Two sets of values were needed to estimate the direct cost: (1) unit cost for each of the identified materials and activities. The unit costs used in this study were actual average costs. (2) Quantity takeoff for each of the materials and activities. For cost estimation purposes, the previously described activities were grouped in some cases. The direct cost consequence of the closure of IH-35E southbound freeway for both cap options (lane closure cost) and the associated indirect cost of impacted economic activity were also considered herein.

For the indirect cost of traffic closure and detour, the following analysis is based on partly site-specific and partly general information. The southbound lanes of IH-35E freeway carry an average daily traffic of 113,676 [32] vehicles of which 8% are single-unit trucks and 4% are combination trucks. A 1.6-km work zone with a 24 h/day four lane closures is assumed herein in effect between Walnut Hill Lane and Royal Lane until the construction is complete. The travel delay cost is a summation of the following costs:Based on the above estimates, the grand total cost due to lane closures on I-35E southbound freeway at the construction site was found as approximately $720,567/day. Gantt charts showing the time schedule of construction are presented in Figs. 16 and 17. The total calculated direct costs were $207,281 and $202,735 for the post-tensioned cap and the composite cap, respectively, as shown in Tables 4 and 5, similar for the two alternatives. However, the lane closure cost of closing IH-35E makes the cast-in-place option much less economic. It may be noted that the construction of the post-tensioned cap would require 23 days, of which traffic would be impacted on 12 days. However, the composite cap would need only 10 days of construction time, of which only one day (actually, a few hours during a night) would have traffic impact. Therefore, the composite alternative had a significant advantage in terms of construction and traffic disruption times. This is apparent in the total traffic disruption costs for the two alternatives, a very high $8.6 million for the post-tensioned cap, but only $720 k for the composite cap. It is apparent that the traffic disruption cost is the major savings aspect for the latter option. The post-tensioned cap option is overall 23 times more expensive than the composite cap option.

Above features an analysis performed for the city of Dallas, Texas, in one of the most heavily populated highways in the metroplex, using a Federal Highway Administration program [41]. The costs associated with lane closures are multi-faceted, and some are difficult to quantify. The main cost factors associated with the traffic impact were included herein. Factors such as noise and its impact on business and local communities are hard to estimate. Another associated factor is emission cost of vehicles. There is no current consensus on how to assign a dollar value to quantify the impacts of each emission pollutant type. Lifetime maintenance cost for the two alternatives were not included herein. Because of the need to periodically restore the corrosion protective coating on the steel cap section, the composite cap would have a higher maintenance cost overall than the post-tensioned cap.

The expected duration of the construction activities was obtained by surveying construction engineers who worked on the cap project. Because of the importance of minimizing traffic disruptions on the major freeways, it was assumed that the work would be performed during all seven days of the week. To estimate the time of IH-35E traffic impact during construction, activities that would impact the traffic were differentiated from those that would not.

The one disadvantage of the proposed composite bent cap is the additional flexibility with large deflections. This requires the fabrication of the steel section with a predefined camber to comply with the serviceability limits.

If different cap types are used for the two pier caps supporting a span, the difference in the deflection developed at each end can generate additional stresses in the superstructure and slab joints. This effect can be amplified for cases like span 7 considered herein, where one of the exterior girders is located near the point of maximum deflection in the cap (Fig. 3).

From the evaluation of the direct costs, construction schedule and traffic impacts, the composite cap was considered a far superior alternative overall.